Allow serving BookWyrm on a non-standard port

.env.example

@@ -16,6 +16,11 @@ DEFAULT_LANGUAGE="English"
 ## Leave unset to allow all hosts
 # ALLOWED_HOSTS="localhost,127.0.0.1,[::1]"
 
+# Specify when the site is served from a port that is not the default
+# for the protocol (80 for HTTP or 443 for HTTPS).
+# Probably only necessary in development.
+# PORT=1333
+
 MEDIA_ROOT=images/
 
 # Database configuration
@@ -139,9 +144,9 @@ HTTP_X_FORWARDED_PROTO=false
 TWO_FACTOR_LOGIN_VALIDITY_WINDOW=2
 TWO_FACTOR_LOGIN_MAX_SECONDS=60
 
-# Additional hosts to allow in the Content-Security-Policy, "self" (should be DOMAIN)
+# Additional hosts to allow in the Content-Security-Policy, "self" (should be
-# and AWS_S3_CUSTOM_DOMAIN (if used) are added by default.
+# DOMAIN with optionally ":" + PORT) and AWS_S3_CUSTOM_DOMAIN (if used) are
-# Value should be a comma-separated list of host names.
+# added by default.  Value should be a comma-separated list of host names.
 CSP_ADDITIONAL_HOSTS=
 
 # Time before being logged out (in seconds)

bookwyrm/connectors/connector_manager.py

@@ -118,9 +118,9 @@ def get_connectors() -> Iterator[abstract_connector.AbstractConnector]:
 def get_or_create_connector(remote_id: str) -> abstract_connector.AbstractConnector:
     """get the connector related to the object's server"""
     url = urlparse(remote_id)
-    identifier = url.netloc
+    identifier = url.hostname
     if not identifier:
-        raise ValueError("Invalid remote id")
+        raise ValueError(f"Invalid remote id: {remote_id}")
 
     try:
         connector_info = models.Connector.objects.get(identifier=identifier)
@@ -188,8 +188,11 @@ def raise_not_valid_url(url: str) -> None:
     if not parsed.scheme in ["http", "https"]:
         raise ConnectorException("Invalid scheme: ", url)
 
+    if not parsed.hostname:
+        raise ConnectorException("Hostname missing: ", url)
+
     try:
-        ipaddress.ip_address(parsed.netloc)
+        ipaddress.ip_address(parsed.hostname)
         raise ConnectorException("Provided url is an IP address: ", url)
     except ValueError:
         # it's not an IP address, which is good

bookwyrm/forms/links.py

@@ -26,7 +26,7 @@ class FileLinkForm(CustomForm):
         url = cleaned_data.get("url")
         filetype = cleaned_data.get("filetype")
         book = cleaned_data.get("book")
-        domain = urlparse(url).netloc
+        domain = urlparse(url).hostname
         if models.LinkDomain.objects.filter(domain=domain).exists():
             status = models.LinkDomain.objects.get(domain=domain).status
             if status == "blocked":

bookwyrm/models/connector.py

@@ -11,7 +11,7 @@ ConnectorFiles = models.TextChoices("ConnectorFiles", CONNECTORS)
 class Connector(BookWyrmModel):
     """book data source connectors"""
 
-    identifier = models.CharField(max_length=255, unique=True)
+    identifier = models.CharField(max_length=255, unique=True)  # domain
     priority = models.IntegerField(default=2)
     name = models.CharField(max_length=255, null=True, blank=True)
     connector_file = models.CharField(max_length=255, choices=ConnectorFiles.choices)

bookwyrm/models/federated_server.py

@@ -16,7 +16,7 @@ FederationStatus = [
 class FederatedServer(BookWyrmModel):
     """store which servers we federate with"""
 
-    server_name = models.CharField(max_length=255, unique=True)
+    server_name = models.CharField(max_length=255, unique=True)  # domain
     status = models.CharField(
         max_length=255, default="federated", choices=FederationStatus
     )
@@ -64,5 +64,4 @@ class FederatedServer(BookWyrmModel):
     def is_blocked(cls, url: str) -> bool:
         """look up if a domain is blocked"""
         url = urlparse(url)
-        domain = url.netloc
+        return cls.objects.filter(server_name=url.hostname, status="blocked").exists()
-        return cls.objects.filter(server_name=domain, status="blocked").exists()

bookwyrm/models/link.py

@@ -38,7 +38,7 @@ class Link(ActivitypubMixin, BookWyrmModel):
         """create a link"""
         # get or create the associated domain
         if not self.domain:
-            domain = urlparse(self.url).netloc
+            domain = urlparse(self.url).hostname
             self.domain, _ = LinkDomain.objects.get_or_create(domain=domain)
 
         # this is never broadcast, the owning model broadcasts an update

bookwyrm/models/user.py

@@ -349,7 +349,7 @@ class User(OrderedCollectionPageMixin, AbstractUser):
         if not self.local and not re.match(regex.FULL_USERNAME, self.username):
             # generate a username that uses the domain (webfinger format)
             actor_parts = urlparse(self.remote_id)
-            self.username = f"{self.username}@{actor_parts.netloc}"
+            self.username = f"{self.username}@{actor_parts.hostname}"
 
         # this user already exists, no need to populate fields
         if not created:
@@ -558,7 +558,7 @@ def set_remote_server(user_id, allow_external_connections=False):
     user = User.objects.get(id=user_id)
     actor_parts = urlparse(user.remote_id)
     federated_server = get_or_create_remote_server(
-        actor_parts.netloc, allow_external_connections=allow_external_connections
+        actor_parts.hostname, allow_external_connections=allow_external_connections
     )
     # if we were unable to find the server, we need to create a new entry for it
     if not federated_server:

bookwyrm/settings.py

@@ -350,28 +350,31 @@ USE_L10N = True
 
 USE_TZ = True
 
-
-USER_AGENT = f"BookWyrm (BookWyrm/{VERSION}; +https://{DOMAIN}/)"
-
 # Imagekit generated thumbnails
 ENABLE_THUMBNAIL_GENERATION = env.bool("ENABLE_THUMBNAIL_GENERATION", False)
 IMAGEKIT_CACHEFILE_DIR = "thumbnails"
 IMAGEKIT_DEFAULT_CACHEFILE_STRATEGY = "bookwyrm.thumbnail_generation.Strategy"
 
-# Static files (CSS, JavaScript, Images)
-# https://docs.djangoproject.com/en/3.2/howto/static-files/
-
 PROJECT_DIR = os.path.dirname(os.path.abspath(__file__))
 CSP_ADDITIONAL_HOSTS = env.list("CSP_ADDITIONAL_HOSTS", [])
 
-# Storage
-
 PROTOCOL = "http"
 if USE_HTTPS:
     PROTOCOL = "https"
     SESSION_COOKIE_SECURE = True
     CSRF_COOKIE_SECURE = True
 
+PORT = env.int("PORT", 443 if USE_HTTPS else 80)
+if (USE_HTTPS and PORT == 443) or (not USE_HTTPS and PORT == 80):
+    NETLOC = DOMAIN
+else:
+    NETLOC = f"{DOMAIN}:{PORT}"
+BASE_URL = f"{PROTOCOL}://{NETLOC}"
+
+USER_AGENT = f"BookWyrm (BookWyrm/{VERSION}; +{BASE_URL})"
+
+# Storage
+
 USE_S3 = env.bool("USE_S3", False)
 USE_AZURE = env.bool("USE_AZURE", False)
 S3_SIGNED_URL_EXPIRY = env.int("S3_SIGNED_URL_EXPIRY", 900)
@@ -440,11 +443,11 @@ elif USE_AZURE:
 else:
     # Static settings
     STATIC_URL = "/static/"
-    STATIC_FULL_URL = f"{PROTOCOL}://{DOMAIN}{STATIC_URL}"
+    STATIC_FULL_URL = BASE_URL + STATIC_URL
     STATICFILES_STORAGE = "django.contrib.staticfiles.storage.StaticFilesStorage"
     # Media settings
     MEDIA_URL = "/images/"
-    MEDIA_FULL_URL = f"{PROTOCOL}://{DOMAIN}{MEDIA_URL}"
+    MEDIA_FULL_URL = BASE_URL + MEDIA_URL
     DEFAULT_FILE_STORAGE = "django.core.files.storage.FileSystemStorage"
     # Exports settings
     EXPORTS_STORAGE = "bookwyrm.storage_backends.ExportsFileStorage"

bookwyrm/templatetags/utilities.py

@@ -120,7 +120,7 @@ def id_to_username(user_id):
     """given an arbitrary remote id, return the username"""
     if user_id:
         url = urlparse(user_id)
-        domain = url.netloc
+        domain = url.hostname
         parts = url.path.split("/")
         name = parts[-1]
         value = f"{name}@{domain}"

bookwyrm/tests/test_signing.py

@@ -15,7 +15,7 @@ from django.utils.http import http_date
 
 from bookwyrm import models
 from bookwyrm.activitypub import Follow
-from bookwyrm.settings import DOMAIN
+from bookwyrm.settings import DOMAIN, NETLOC
 from bookwyrm.signatures import create_key_pair, make_signature, make_digest
 
 
@@ -77,7 +77,7 @@ class Signature(TestCase):
                 "HTTP_SIGNATURE": signature,
                 "HTTP_DIGEST": digest,
                 "HTTP_CONTENT_TYPE": "application/activity+json; charset=utf-8",
-                "HTTP_HOST": DOMAIN,
+                "HTTP_HOST": NETLOC,
             },
         )
 

pytest.ini

@@ -11,6 +11,7 @@ env =
     DEBUG = false
     USE_HTTPS = true
     DOMAIN = your.domain.here
+    PORT = 4242
     ALLOWED_HOSTS = your.domain.here
     BOOKWYRM_DATABASE_BACKEND = postgres
     MEDIA_ROOT = images/